A Selective Peroxisome Proliferator-Activated Receptor ␦ Agonist Promotes Reverse Cholesterol Transport

A selective peroxisome proliferator-activated receptor ␦ agonist promotes reverse cholesterol transport

William R. Oliver, Jr.*, Jennifer L. Shenk†, Mike R. Snaith†‡, Caroline S. Russell‡, Kelli D. Plunket†, Noni L. Bodkin§, Michael C. Lewis*, Deborah A. Winegar*, Marcos L. Sznaidman†, Millard H. Lambert†, H. Eric Xu†, Daniel D. Sternbach*†, Steven A. Kliewer†, Barbara C. Hansen§, and Timothy M. Willson†¶

*Metabolic Diseases Drug Discovery and †Nuclear Receptor Discovery Research, GlaxoSmithKline, Research Triangle Park, NC 27709; ‡Cardiovascular Diseases Drug Discovery, GlaxoSmithKline, Stevenage, Hertfordshire SG1 2NY, United Kingdom; and §Obesity and Diabetes Research Center, University of Maryland School of Medicine, Baltimore, MD 21201

Edited by Keith R. Yamamoto, University of California, San Francisco, CA, and approved February 26, 2001 (received for review January 12, 2001) The peroxisome proliferator-activated receptors (PPARs) are di- ripheral cells, including the macrophage-derived foam cells, and etary lipid sensors that regulate fatty acid and carbohydrate returned to the liver (9). Agents that raise the levels of HDL metabolism. The hypolipidemic effects of the fibrate drugs and the through reverse cholesterol transport could provide a new antidiabetic effects of the glitazone drugs in humans are due to therapeutic option for the prevention of atherosclerotic cardio- activation of the ␣ (NR1C1) and ␥ (NR1C3) subtypes, respectively. vascular disease (10). By contrast, the therapeutic potential of the ␦ (NR1C2) subtype is Recently, the ATP-binding cassette A1 (ABCA1) protein has unknown, due in part to the lack of selective ligands. We have used been identified as a regulator of cholesterol and phospholipid combinatorial chemistry and structure-based drug design to de- transport from cells (11). Patients with Tangier disease or velop a potent and subtype-selective PPAR␦ agonist, GW501516. In familial hypoalphalipoproteinemia have been identified with macrophages, fibroblasts, and intestinal cells, GW501516 increases loss-of-function mutations in the ABCA1 gene (12–14). These expression of the reverse cholesterol transporter ATP-binding patients have low levels of HDLc and high triglycerides and show cassette A1 and induces apolipoprotein A1-specific cholesterol an increased incidence of cardiovascular disease (15). Thus, efflux. When dosed to insulin-resistant middle-aged obese rhesus therapies that increase the expression of ABCA1 could provide monkeys, GW501516 causes a dramatic dose-dependent rise in a new approach to treating atherogenic dyslipidemia (9). RXR serum high density lipoprotein cholesterol while lowering the heterodimers may play a role in the transcriptional regulation of levels of small-dense low density lipoprotein, fasting triglycerides, ABCA1 expression, because RXR agonists have been shown to and fasting insulin. Our results suggest that PPAR␦ agonists may be increase the expression of ABCA1 in the intestine of mice (16). effective drugs to increase reverse cholesterol transport and de- In this report, we demonstrate that a selective PPAR␦ agonist crease cardiovascular disease associated with the metabolic syn- increases ABCA1 expression and cholesterol efflux from cells drome X. and increases HDLc in primates. PPAR␦ agonists may provide a new approach to the treatment of cardiovascular disease by promoting reverse cholesterol transport. he peroxisome-proliferator activated receptors (PPARs) are Tmembers of the nuclear receptor gene family that are Materials and Methods activated by fatty acids and fatty acid metabolites (1, 2). The Reagents and Assays. The human PPAR␦ binding assay was PPARs belong to the subset of nuclear receptors that function performed as described using [3H]GW2433 as the radioligand (2, as heterodimers with the 9-cis-retinoic acid receptor (RXR) (3). 17). Expression plasmids for the nuclear receptor-GAL4 chime- Three subtypes, designated PPAR␣ (NR1C1), PPAR␥ ␦ ras were prepared by inserting amplified cDNAs encoding the (NR1C3), and PPAR (NR1C2), are found in species ranging ligand binding domains into a modified pSG5 expression vector from Xenopus to humans (1). Each receptor has a distinct tissue (Stratagene) containing the GAL4 DNA-binding domain (ami- expression profile, with PPAR␣ the main subtype in the liver, ␥ ␦ no acids 1–147) and the simian virus 40 large T antigen nuclear PPAR the main subtype in adipose tissue, and PPAR ex- localization signal (APKKKRKVG). Transient transfection pressed in many tissues (4). The physiological functions of ␣ ␥ assays were performed as described using (UAS)5-tk-luciferase PPAR and PPAR in lipid and carbohydrate metabolism were reporter constructs (18). The synthesis of (2-methyl-4- uncovered once it was recognized that they were the receptors (((4-methyl-2-(4-trifluoromethylphenyl)-1,3-thiazol-5-yl)- for the fibrate and glitazone drugs, respectively (1). By contrast, ␦ methyl)sulfanyl)phenoxy)acetic acid (GW501516), 2-(4-(2-(1- PPAR (NR1C2) is not reported to be a receptor for any known cyclohexanebutyl-3-cyclohexylureido)ethyl)phenylthio)-2- class of drug molecules, and its role in mammalian physiology methylpropionic acid (GW7647), and 3-(3-(2-chloro-3- has remained undefined (1). trifluoromethylbenzyl-2,2-diphenylethylamino)propoxy) In humans, lipid homestasis is a delicate balance between phenylacetic acid (GW3965) will be described elsewhere. dietary intake, de novo synthesis, and catabolism. The increased incidence of cardiovascular disease in Westernized nations has Cell Culture. THP1 human monocyte cells (ATCC TID-202) were been linked to dyslipidemias associated with changes in the fat cultured in RPMI 1640 medium containing 10% FBS. Differ- content of the diet (5). Obesity, insulin resistance, and hypertension are comorbidities with these lipid disorders, which together are known as the metabolic syndrome X (6). Individuals This paper was submitted directly (Track II) to the PNAS office. with this condition have raised serum triglycerides and abnor- Abbreviations: PPAR, peroxisome proliferator-activated receptor; RXR, 9-cis-retinoic acid mally low levels of high density lipoprotein cholesterol (HDLc) receptor; HDL, high density lipoprotein; HDLc, HDL cholesterol; ABC, ATP binding cassette; (6, 7). This lipid profile is accompanied by an increase in the LDL, low density lipoprotein; VLDL, very low density lipoprotein; apo, apolipoprotein; LXR, liver X receptor. proportion of small-dense low density lipoprotein (LDL) parti- ¶To whom reprint requests should be addressed at: GlaxoSmithKline, Five Moore Drive, cles, which are prone to accumulate in the arterial wall leading NTH-M1421, Research Triangle Park, NC 27709-3398. E-mail: [email protected]. to the formation of atherosclerotic cholesterol-laden foam cells The publication costs of this article were defrayed in part by page charge payment. This (8). HDL plays a protective role through the process of reverse article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. cholesterol transport whereby cholesterol is removed from pe- §1734 solely to indicate this fact.

5306–5311 ͉ PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 www.pnas.org͞cgi͞doi͞10.1073͞pnas.091021198 Downloaded by guest on September 28, 2021 entiation into macrophages was performed on 6-well plates (1 ϫ below 30 mg͞dl (B.C.H., unpublished work). One animal, whose 106 cells͞well) by treatment with phorbol 12-myristate 13- triglycerides were Ͻ20 mg͞dl and HDLc was Ͼ130 mg͞dl at the acetate (100 ng͞ml) for 5 days. Phorbol 12-myristate 13-acetate 1.0 mg͞kg dose, showed a reduction in food intake. As a (100 ng͞ml) was included in the medium of all subsequent precaution, this animal was removed from the study before experiments with these cells to maintain differentiation. 1BR3N administration of the 3.0 mg͞kg dose. Blood samples were drawn human skin fibroblasts (European Collection of Cell Cultures under light ketamine sedation on day 28 of each dose after a 16 h 90020508) and FHS74 human intestinal cells (ATCC CCL-241) overnight fast and2hafterdosing. At the completion of the were cultured in their recommended maintenance media and study the rhesus monkeys were monitored through an 8-week plated on 6-well plates (1 ϫ 106 cells͞well) before the efflux and washout period. Studies were approved by the Institutional gene expression studies. Animal Care and Use committees for the University of Mary- land and GlaxoSmithKline. Cholesterol Efflux and Gene Expression Studies. Phorbol 12- myristate 13-acetate-differentiated THP1 macrophages were Serum Analysis. Serum concentrations of GW501516 were deter- washed in serum-free medium and incubated in 5% FBS medium mined by HPLC͞MS in the Department of BioMedical Analysis containing 1 ␮Ci͞ml of [3H]cholesterol (New England Nuclear) at GlaxoSmithKline. Serum total cholesterol, HDLc, triglycer- and 1.5% fatty acid-free BSA for 24 h. The cells were washed in ides, and glucose were determined by an Instrumentation Lab- serum-free medium and incubated for 24 h in serum-free oratory (Lexington, MA) IL600 clinical chemistry analyzer. medium supplemented with 1.5% fatty acid-free BSA and test Serum insulin was determined by immunoassay (Linco Research compound or vehicle (0.1% DMSO). The equilibration medium Immunoassay, St. Charles, MO). Serum apoAI, apoAII, and was removed and the cells were washed twice in serum-free apoCIII were determined by immunoprecipitation (Penn Med- medium. Serum-free RPMI 1640 medium containing test com- ical Laboratory, Washington, DC). pound or vehicle Ϯ purified human apolipoprotein (apo) AI (Athens Research & Technology, Athens, GA) was added. The Lipoprotein Subclass Analysis. Subclass analysis was performed by final concentration of apoA1 was 10 ␮g͞ml. The cells were proton NMR spectroscopy (LipoMed, Raleigh, NC) on EDTA maintained for 24 h before medium and cell extracts were plasma samples as described (20). The lipoprotein subclasses isolated. Scintillation counting was performed to determine the were defined by NMR-determined particle size as follows: small percentage cellular cholesterol efflux Ϯ apoAI. Cholesterol HDL, 7.3–8.8 nm; large HDL, 8.8–13 nm; small LDL, 18.8–19.7 efflux studies with the 1BR3N fibroblasts were performed as nm; medium LDL, 19.8–21.2 nm; large LDL, 21.3–23.0 nm; small described for the THP1 cells, except that cells were maintained very low density lipoprotein (VLDL), 27–35 nm; medium for 48 h before scintillation counting was performed. VLDL, 35–60 nm; and large VLDL, 60–200 nm. Parallel nonradioactive experiments were performed with the THP1 macrophages and 1BR3N fibroblasts to generate samples Statistical Analyses. Unless otherwise stated, data are expressed for quantitation of ABCA1, PPAR␦, and liver X receptor ␣ as mean Ϯ SE. The significance of differences were analyzed by (LXR␣) mRNA. Gene expression data with the FHS74 intesti- using a paired Student’s t test. nal cells was generated after treatment with test compound or

vehicle for 48 h. Total RNA was generated by using the Qiagen Results PHARMACOLOGY (Chatsworth, CA) RNeasy Mini Kit, and DNase was treated Development of a Selective PPAR␦ Agonist. To evaluate the thera- according to the manufacturer’s protocol (Ambion). mRNA peutic potential of PPAR␦ agonists, we developed a subtype- expression was analyzed by using real-time quantitative-PCR on selective small molecule ligand by using combinatorial chemistry an Applied Biosystems Prism 7700 sequence detection system. (ref. 17; D.D.S., unpublished results) and structure-based drug Primer͞probe sequences used were as follows: ABCA1 forward design (ref. 2; H.E.X. and M.H.L., unpublished results). primer, 5Ј-TGTCCAGTCCAGTAATGGTTCTGTGT-3Ј, re- GW501516 (Fig. 1A) is a high affinity ligand in a human PPAR␦ Ј Ј ϭ Ϯ verse primer 5 -GCGAGATATGGTCCGGATTG-3 , probe, binding assay (2) with Ki 1.1 0.1 nM (data not shown). When 5ЈFAM-ACACCTGGAGAGAAGCTTTCAACGAGAC- evaluated in a cell-based transfection assay, using the ligand TAACC-TAMRA3Ј; PPAR␦ forward primer, 5Ј-CAACTGCA- binding domain of human PPAR␦ fused to the DNA-binding GATGGGCTGTGA-3Ј, reverse primer, 5Ј-ATGCATGAA- domain of the yeast transcription factor GAL4, GW501516 CACCGTAGTGGAA-3Ј, probe, 5ЈFAM-CCCGGCACT- induced expression of a GAL4-responsive reporter gene with an Ј ␣ Ј ϭ Ϯ Ͼ CCATGTTGAGGCT-TAMRA3 ; LXR forward primer, 5 - EC50 1.2 0.1 nM. GW501516 was 1,000-fold selective for AGCCCTGCATGCCTACGTC-3Ј, reverse primer, 5Ј-GCATC- PPAR␦ over the other subtypes (Fig. 1B). At doses up to 1.0 ␮M, CGTGGGAACATCAG-3Ј, probe, 5ЈFAM-CCATCCAC- GW501516 did not promote adipocyte differentiation, bind to CATCCCCATGACC G-TAMRA3Ј. Expression data were RXR␣, or have activity on other nuclear or non-nuclear recep- normalized to 18S as described by the vendor. tors (Fig. 1C and data not shown). GW501516 also shows EC50 ϭ Ϯ ␦ ϭ Ϯ 1.6 0.3 nM on rhesus PPAR and EC50 24 2nMon Primate Study. GW501516 was evaluated in six male obese rhesus mouse PPAR␦ with Ͼ500-fold selectivity over the other subtypes monkeys from the colony at the Obesity and Diabetes Research (data not shown). Other PPAR␦ agonists lack selectivity over Center at the University of Maryland (19). The animals were either PPAR␣ [e.g., GW2433 (1, 17)] or PPAR␥ [e.g., L-165041 housed with ad libitum access to food and water. Caloric (1, 21)] or have activity on eicosanoid receptors [e.g., carbo- composition of the diet was 13% fat, 18% protein, and 69% prostacyclin (22)]. digestible carbohydrates. Each animal received repeat oral administration of ␣-tocopheryl polyethylene glycol succinate͞ PPAR␦ Regulates ABCA1 and Cholesterol Efflux. Recently, several polyethylene glycol 400 (vehicle) or GW501516, formulated as a nuclear receptors that form heterodimers with RXR have been solution of the sodium salt in vehicle, delivered in small food evaluated for their ability to regulate expression of the choles- snacks twice a day. After 4 weeks of dosing with vehicle, the terol transporter ABCA1 (16). Ligands for either RXR or LXR␣ initial dose of GW501516 was 0.1 mg͞kg for 4 weeks, rising to 0.3, induce ABCA1 expression in murine peritoneal macrophages, 1.0, and 3.0 mg͞kg during sequential 4-week periods. Food human primary macrophages, or THP1 macrophages (16, 23, consumption was monitored daily, and body weights were re- 24). To assess the potential for the PPARs to regulate this corded weekly. Decreases in food consumption have been process, THP1 macrophages were dosed with ligands selective observed in these primates with agents that lower triglycerides for each of the three subtypes. At 100 nM, GW7647 (see

Oliver et al. PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 ͉ 5307 Downloaded by guest on September 28, 2021 Fig. 2. Regulation of ABCA1 expression and cholesterol efflux from THP1 macrophages. Compounds were used at the following concentrations: LXR␣ (GW3965), 1.0 ␮M; PPAR␦ (GW501516), 100 nM; PPAR␣ (GW7647), 100 nM; PPAR␥ (GW7845), 100 nM. Data are presented as the mean of assays performed in triplicate Ϯ SD. (A) ABCA1 mRNA levels. (B) ApoA1-specific cholesterol efflux.

induction of ABCA1 mRNA expression (Fig. 2A). The PPAR␥ agonist GW7845 also induced ABCA1 expression, whereas the PPAR␣ agonist GW7647 showed only a weak effect. In parallel with the measurement of ABCA1 expression, we monitored the ability of the selective PPAR agonists to effect cholesterol efflux from the THP1 macrophages (Fig. 2B). Although not as effective as the LXR␣ agonist, the PPAR␦ agonist GW501516 produced a 2-fold increase in cholesterol efflux to apoAI (Fig. 2B and Table 1). Thus, PPAR␦ is an RXR partner that regulates cholesterol efflux from macrophages. Both the PPAR␣ and PPAR␥ agonists were inactive despite their ability to produce small increases in ABCA1 expression, which suggests that activation of pathways in addition to ABCA1 may be required to increase cholesterol efflux from these cells. Fig. 1. GW501516 is a selective PPAR␦ agonist. Data are expressed as fold PPAR␦ is expressed in many tissues that contribute to cho- change compared with vehicle-treated cells and represent the mean of assays lesterol flux (4). To evaluate whether PPAR␦ plays a broad role performed in triplicate Ϯ SE. (A) Chemical structure of GW501516. (B) Acti- in the regulation of reverse cholesterol transport, we evaluated vation of GAL4-PPAR ligand binding domain chimeras in transiently trans- GW501516 in human fibroblast and intestinal cell lines in fected CV-1 cells by GW501516. Dose–response curves are shown for human ␦ F ␣ ‚ ␥ ᮀ addition to THP1 macrophages (Table 1). In 1BR3N human skin PPAR ( ), human PPAR ( ), and human PPAR ( ). (C) Activation of GAL4- fibroblasts, the PPAR␦ agonist produced a 3.4-fold increase in nuclear receptor ligand binding domain chimeras in CV-1 cells. Transfected cells were treated with 1.0 ␮M GW501516. ABCA1 expression and a 2-fold increase in apoA1-specific cholesterol efflux. In FHS74 human intestinal cells (26), the PPAR␦ agonist produced a 2.1-fold increase in ABCA1 expression. Little or no changes in the expression of LXR␣ (Table 1) Materials and Methods; P. J. Brown, personal communication) is ␦ a selective PPAR␣ agonist (EC ϭ 6, 1,070, and 6,200 nM on and no changes in the expression of PPAR (data not shown) 50 were observed in the macrophages, fibroblasts, or intestinal cells. human PPAR␣, PPAR␥, and PPAR␦, respectively), GW7845 (1, ␦ ␥ ϭ Ͼ These results suggest that PPAR agonists may promote cho- 25) is a selective PPAR agonist (EC50 3,500, 0.7, and 10,000 ␣ ␥ ␦ lesterol efflux from multiple tissues by increasing expression of nM on human PPAR , PPAR , and PPAR , respectively), and the reverse cholesterol transporter ABCA1. GW501516 is a selective PPAR␦ agonist (Fig. 1B). Compared with a synthetic LXR␣ agonist GW3965 (see Materials and A PPAR␦ Agonist Raises HDLc in a Primate Model of the Metabolic ␣ ϭ Methods; LXR EC50 190 nM; J. L. Collins, personal com- Syndrome X. To assess whether the activity of GW501516 on munication), the PPAR␦ agonist GW501516 showed strong ABCA1 expression and cholesterol efflux in cells would trans-

Table 1. Effect of GW501516 on cholesterol efﬂux and mRNA expression Cell line Cell type Cholesterol efﬂux, % ABCA1 mRNA, % LXR␣ mRNA, %

THP1 Macrophage 200 Ϯ 15 560 Ϯ 150 190 Ϯ 40 1BR3N Fibroblast 210 Ϯ 10 340 Ϯ 25 115 Ϯ 20 FHS74 Intestinal N.D. 215 Ϯ 30 120 Ϯ 10

Data are presented as the mean of the percent change (%) relative to vehicle from assays performed in triplicate Ϯ SD. Vehicle ϭ 100%. N.D. ϭ not determined.

5308 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.091021198 Oliver et al. Downloaded by guest on September 28, 2021 Table 2. Effect of GW501516 on clinical chemistries in male obese rhesus monkeys Dose, mg͞kg, bid Serum conc., ng͞ml Total chol., mg͞dl HDLc, mg͞dl Trig., mg͞dl Insulin, ␮U͞ml Glucose, mg͞dl

0 — 125 Ϯ 10 61 Ϯ 9 132 Ϯ 15 105 Ϯ 24 69 Ϯ 2 0.1 40 Ϯ 91** 135 Ϯ 675Ϯ 9 128 Ϯ 21 113 Ϯ 29 66 Ϯ 2 0.3 71 Ϯ 24** 141 Ϯ 984Ϯ 9* 85 Ϯ 13* 82 Ϯ 19 66 Ϯ 2 1.0 265 Ϯ 100** 151 Ϯ 12 95 Ϯ 11* 66 Ϯ 19* 76 Ϯ 23 71 Ϯ 3 3.0 706 Ϯ 109** 159 Ϯ 10* 109 Ϯ 9** 58 Ϯ 16** 55 Ϯ 10* 68 Ϯ 2 Nonobese — 152 90 48 25 65

Data are presented as mean Ϯ SE for n ϭ 6 for all doses except 3.0 mg͞kg where n ϭ 5. *, P Յ 0.05; **, P Յ 0.01 compared to vehicle treatment. Historical data from lean rhesus monkeys in the same colony are labeled nonobese.

late to a therapeutic benefit, a primate model of human meta- compares well with the conventional enzymatic determination of bolic disease was selected. We previously have described a serum HDLc levels (Table 2). In parallel, we analyzed serum to colony of rhesus monkeys that develop spontaneous adult-onset determine the effect of GW501516 on the apo composition. obesity on standard low-fat diets and show a high risk of ApoAI and apoAII, the major lipoprotein components of HDL developing diabetes, ultimately requiring insulin therapy to (34), were raised by 43% and 21%, respectively, at the 3.0 mg͞kg maintain glycemic control (19). In the prediabetic state these dose of GW501516 (Fig. 4). The increases in apoAI and apoAII primates display many of the features of the human metabolic suggest that the rise in serum HDLc is associated with the syndrome X, including dyslipidemia, insulin resistance, central production of more lipoprotein particles (34). Lipoprotein sub- obesity, hyperinsulinemia, and hypertension (27, 28). Impor- class analysis (20) (Fig. 3B) confirmed that there was little tantly, the dyslipidemia observed in these primates is manifest as change in the relative proportions of the HDL subclasses. Thus, low HDLc and high triglycerides. Treatment of these primates ␣ ␥ the rise in HDLc produced by GW501516 may result from an with either PPAR agonists or PPAR agonists leads to a increase in the number rather than the size of HDL particles. clinical response that parallels the known effects of fibrates and Consistent with the decrease in fasting triglycerides (Table 2), glitazones in humans (19, 29). GW501516 produced a 50% decrease in VLDL (Fig. 3A). Six middle-aged obese rhesus monkeys (age ϭ 12.9 Ϯ 1.5 y, Remarkably, apoCIII, a component of VLDL and HDL that may weight ϭ 16.6 Ϯ 1.7 kg, and rhesus body mass index ϭ 53 Ϯ 5 kg͞m2) were dosed with vehicle for 4 weeks to establish baseline characteristics (Table 2). Compared with their lean counter- parts, the obese rhesus monkeys had low serum HDLc, high fasting triglycerides, and high fasting insulin levels, but were normoglycemic, which are characteristics of prediabetic insulin resistance as defined by longitudinal studies in this population (19, 30). Each animal received increasing doses of GW501516, PHARMACOLOGY with the dose maintained at each level for a 4-week period. A dose-dependent increase in the serum level of GW501516 was detected, reaching Ϸ700 ng͞ml at the highest dose (Table 2). Because Ͼ99% of the circulating GW501516 is bound to plasma proteins (data not shown), we expect activation of only PPAR␦ in vivo at these serum drug concentrations (31, 32). GW501516 had a dramatic effect on the serum lipid profile (Table 2). HDLc increased in a dose-dependent manner over the treatment period, with a 79% increase in HDLc relative to the baseline level at the 3.0 mg͞kg dose. The ratio of HDLc͞total cholesterol also was improved at all doses of GW501516. In parallel with the increase in HDLc, GW501516 produced a dose-dependent lowering of fasting triglycerides, with a 56% decrease at the 3.0 mg͞kg dose. No changes in body weight or liver enzymes were seen (data not shown). At the end of the study the animals were monitored through a washout period, during which serum lipid parameters returned to their baseline values (data not shown).

A PPAR␦ Agonist Produces a Less Atherogenic Lipid Profile. Patients with low HDLc and high triglycerides often show an atherogenic lipid composition containing high levels of small-dense LDL particles (6, 7). To obtain additional characterization of the lipid particle composition in the obese rhesus monkeys after GW501516 therapy, we analyzed plasma samples by proton NMR spectroscopy (33). In humans and obese rhesus monkeys, Fig. 3. NMR analysis of serum lipoproteins from primates treated with GW501516. V ϭ vehicle, GW ϭ 3.0 mg͞kg GW501516. Data are presented as lipoprotein subclasses distinguished by NMR correlate well with mean Ϯ SE for n ϭ 5. , P Յ 0.05; , P Յ 0.01. (A) Lipid values determined by those determined by other methods (20, 33), including ultracen- * ** ͞ summation of the directly measured concentrations of HDL, LDL, and VLDL trifugation (D.A.W., unpublished work). At the 3.0 mg kg dose, subclasses after conversion to cholesterol (c) or triglyceride (TG) units. (B) NMR analysis showed that GW501516 produced an 80% in- Number of LDL particles calculated from the LDL subclasses. (C) Relative crease in HDLc relative to the baseline level (Fig. 3A), which compositions of the HDL, LDL, and VLDL subclasses.

Oliver et al. PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 ͉ 5309 Downloaded by guest on September 28, 2021 play a role in the regulation of peripheral lipoprotein lipase activity (35), was raised by 46% in the drug-treated primates (Fig. 4). This last result is surprising because the triglyceride- lowering activity of fibrates is accompanied by a reduction in the serum levels of apoCIII and an increase in the clearance of VLDL (36). For example, in this population of obese rhesus monkeys, fenofibrate lowered apoCIII by 29% when serum triglycerides were reduced by 55% and HDLc was raised by 35% (19). Thus, the PPAR␦ agonist GW501516 and the PPAR␣ agonist fenofibrate produce opposite effects on serum apoCIII levels, which is an indication that they regulate lipid metabolism by different biochemical mechanisms. NMR analysis revealed a 29% decrease in LDL cholesterol and a 36% decrease in the number of LDL particles in the drug-treated primates (Fig. 3 A and B). Although these changes ϭ are not as dramatic as those in the HDL fraction, further analysis Fig. 4. Serum apo concentrations determined by immunoprecipitation. V vehicle, GW ϭ 3.0 mg͞kg GW501516. Data are presented as mean Ϯ SE for n ϭ (Fig. 3C) showed that GW501516 produced a reduction in the 5. *, P Յ 0.05; **, P Յ 0.01. proportion of small and medium LDL subclasses, while increasing the proportion of large LDL and medium VLDL subclasses. Therefore, the reduction in LDL is due largely to a decrease in closely matches that seen in humans (19). In these primates, the number of small-dense particles (6). Overall, our analysis by GW501516 has beneficial effects on multiple cardiovascular risk three independent methods indicates that GW501516 changes factors, including lipoprotein size and composition, resulting in the serum lipoprotein composition by increasing the number of a potentially less atherogenic lipid profile (6). These changes, HDL particles while producing fewer, yet larger, LDL and which include a marked increase in HDLc, are consistent with an VLDL particles. Together, these effects should result in a less increased flux of cholesterol from peripheral tissues to nascent atherogenic lipid composition. HDL particles. Patients with Tangier disease and familial hy- ␦ poalphalipoproteinia have low circulating levels of HDLc and A PPAR Agonist Corrects Hyperinsulinemia in Primates. The rhesus high triglycerides due to mutations in the ABCA1 gene (11). monkeys in this study were hyperinsulinemic but maintained Fibroblasts from these patients show a reduced capacity for glucose concentrations in the normal range. Although not a cholesterol efflux, which correlates with the decrease in HDLc primary endpoint of the in vivo study, we noted that GW501516 and an increased risk of cardiovascular disease (15). Because, produced a dose-dependent decrease in the serum insulin levels GW501516 increases ABCA1 expression, promotes cholesterol (Table 2). At the 3.0 mg͞kg dose, a 48% decrease in serum efflux from peripheral cell types, and raises HDLc in primates, insulin was seen relative to the baseline level, whereas no changes it appears that activation of PPAR␦ provides a novel mechanism in fasting glucose levels were detected throughout the study. for promoting reverse cholesterol transport (9). Additional Thus, GW501516 partially corrects the hyperinsulinemia in these studies will be required, however, to determine the global effect primates without adverse effects on glycemic control. The mech- of GW501516 on cholesterol flux from peripheral tissues to the anism by which PPAR␦ lowers insulin in these primates remains liver. to be determined. Fibrates are a class of drugs that have been used for decades Discussion for their beneficial effects on serum lipids. Although fibrates are During the past decade the role of the orphan nuclear receptors predominantly triglyceride-lowering drugs that only modestly PPAR␣ and PPAR␥ in the regulation of lipid metabolism has raise HDLc (19, 36), clinical trials have shown that they lower the been clearly established through their association with the incidence of atherosclerosis and coronary artery disease in fibrate and glitazone drugs, respectively (1). There have been patients with normal levels of LDLc (36, 41). Most fibrate drugs ␦ are only weakly active on human PPAR␣ and show low selec- several tantalizing clues that PPAR also may modulate aspects ␦ ␥ of lipid homeostasis: PPAR␦ binds to many of the same fatty tivity over human PPAR and PPAR (1). It was recently acids as the other subtypes (2), signifying that it also may be a reported that the experimental fibrate drug Wy14,643 induces ␦ ABCA1 expression and cholesterol efflux from macrophages dietary lipid sensor; L-165041, a weak nonselective PPAR ␮ agonist (1), raised total cholesterol levels in db͞db mice (37); and (42). However, at the concentrations used in the study (50 M), ␦ ϭ ␮ PPAR␦ null mice displayed a reduction in the size of adipose Wy14,643 has significant PPAR activity [EC50 35 M for ␦ tissue depots (38), although no lipid phenotype was reported. human PPAR (1)]. Using compounds that are selective for each However, in the absence of potent and selective ligands, the of the three PPAR subtypes, we have now shown that their ␦ Ͼ physiological role of PPAR␦ has remained an enigma. In this relative ability to induce ABCA1 expression is PPAR ␥ Ͼ ␣ report, we have described a truly selective PPAR␦ agonist, PPAR PPAR . These data argue that the reported effects GW501516. Using GW510516 as a chemical tool, we have (42) of high doses of fibrates on cholesterol efflux are mediated provided evidence that PPAR␦ increases cholesterol efflux from primarily through PPAR␦. Glitazone PPAR␥ agonists also were cells, in part, through an increase in the expression of the reported to increase ABCA1 expression through the induction ABCA1 reverse cholesterol transporter. These data suggest that of LXR␣ expression (42, 43). Interestingly, we observed no PPAR␦ is an important regulator of cholesterol metabolism with consistent increase in LXR␣ expression with GW501516 (Table unique pharmacology that distinguishes it from the other PPAR 1) suggesting that alternate mechanisms may contribute to the subtypes. regulation of ABCA1 expression by PPAR␦ agonists. There are significant differences in the regulation of lipid The therapeutic effects of fibrates are due, in part, to their metabolism in rodents compared with humans (39). This is effects on hepatic gene expression mediated by PPAR␣ (36). In highlighted by the fact that there is no single rodent model of particular, decreases in the expression of apoCIII have been dyslipidemia in which both PPAR␣ and PPAR␥ agonists are associated with the triglyceride-lowering activity of PPAR␣ active (40). We chose to study the pharmacology of a PPAR␦ agonists (44–46). We were surprised to find that GW501516 ligand in obese rhesus monkeys where the lipid profile more raised serum apoCIII levels in the obese rhesus monkeys.

5310 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.091021198 Oliver et al. Downloaded by guest on September 28, 2021 Although both PPAR␣ and PPAR␦ agonists lower triglycerides agonists are likely to have beneficial effects on the lipid triad in primates, GW501516 has a greater effect on HDLc compared and the atherogenic particle composition through a mecha- to a fibrate. It is possible that the observed increase in apoCIII nism that increases cholesterol flux from peripheral tissues. after activation of PPAR␦ is due to an increase in the number These activities, combined with the benefit of lowering serum of these apos associated with the HDL particles. In addition, insulin levels, suggest that PPAR␦ agonists may be powerful GW501516 does not affect apoA1 or apoCIII mRNA levels in drugs for decreasing the incidence of cardiovascular disease human hepatocytes (data not shown), providing further evidence associated with the metabolic syndrome X. that changes in hepatic gene expression are unlikely to explain the changes in serum HDLc or triglycerides. We thank Lisa Leesnitzer for PPAR␦ binding data, Jon Collins for Hyperinsulinemia and the lipid triad of low HDLc, small unpublished data on GW3965, Peter Brown for unpublished data on LDL particles, and elevated serum triglycerides are charac- GW7647, Annette Graham (Royal Free Hospital, London) for assistance in developing the cholesterol efflux assay, Theresa Alexander, teristics of dyslipidemia associated with the metabolic syn- Michelle Izuka, Wallace Evans, and Karen Brocklehurst for assistance drome X (6, 7). Individuals with this atherogenic lipoprotein with the primate study, and Jane Binz for analysis of clinical chemistries. phenotype have a higher incidence of premature coronary We also thank Henry Ginsberg (Columbia University), David Hassall, artery disease (6). Our results demonstrate that PPAR␦ and Jeff Cobb for comments on the manuscript.

1. Willson, T. M., Brown, P. J., Sternbach, D. D. & Henke, B. R. (2000) J. Med. 24. Costet, P., Luo, Y., Wang, N. & Tall, A. R. (2000) J. Biol. Chem. 275, Chem. 43, 527–550. 28240–28245. 2. Xu, H. E., Lambert, M. H., Montana, V. G., Parks, D. J., Blanchard, S. G., 25. Cobb, J. E., Blanchard, S. G., Boswell, E. G., Brown, K. K., Charifson, P. S., Brown, P. J., Sternbach, D. D., Lehmann, J. M., Wisely, G. B., Willson, T. M., Cooper, J. P., Collins, J. L., Dezube, M., Henke, B. R., Hull-Ryde, E. A., et al. et al. (1999) Mol. Cell. 3, 397–403. (1998) J. Med. Chem. 41, 5055–5069. 3. Mangelsdorf, D. J. & Evans, R. M. (1995) Cell 83, 841–850. 26. Owens, R. B., Smith, H. S., Nelson-Rees, W. A. & Springer, E. L. (1976) J. Natl. 4. Braissant, O., Foufelle, F., Scotto, C., Dauca, M. & Wahli, W. (1996) Cancer Inst. 56, 843–849. Endocrinology 137, 354–366. 27. Bodkin, N. L., Metzger, B. L. & Hansen, B. C. (1989) Am. J. Physiol. 256, 5. Lichtenstein, A. H., Kennedy, E., Barrier, P., Ernst, N. D., Grundy, S. M., E676–E681. Leveille, G. A., Van Horne, L., Williams, C. L. & Booth, S. L. (1998) Nutr. Rev. 28. Hannah, J. S., Verdery, R. B., Bodkin, N. L., Hansen, B. C., Le, N.-A. & 56, S3–S28. Howard, B. V. (1991) J. Clin. Endocrinol. 72, 1067–1072. 6. Grundy, S. M. (1998) Am. J. Cardiol. 81, 18B–25B. 29. Kemnitz, J. W., Elson, D. F., Roecker, E. B., Baum, S. T., Bergman, R. N. & 7. Ginsberg, H. N. (2000) J. Clin. Invest. 106, 453–458. Meglasson, M. D. (1994) Diabetes 43, 204–211. 8. Marais, A. D. (2000) Curr. Opin. Lipidol. 11, 597–602. 30. Hansen, B. C. & Bodkin, N. L. (1986) Diabetologia 29, 713–719. 9. Tall, A. R. & Wang, N. (2000) J. Clin. Invest. 106, 1205–1207. 10. von Eckardstein, A. & Assmann, G. (2000) Curr. Opin. Lipidol. 11, 627–637. 31. Young, M. A., Eckland, D. J. A., Eastmond, R. & Lettis, S. (1998) Ann. Med. 11. Schmitz, G., Kaminski, W. E. & Orso, E. (2000) Curr. Opin. Lipidol. 11, 493–501. 30, 206–212. 12. Brooks-Wilson, A., Marcil, M., Clee, S. M., Zhang, L.-H., Roomp, K., Van 32. Loi, C. M., Young, M., Randinitis, E., Vassos, A. & Koup, J. P. (1999) Clin. Dam, M., Yu, L., Brewer, C., Collins, J. A., Molhuizen, H. O. F., et al. (1999) Pharmacokinet. 37, 91–104. Nat. Genet. 22, 336–345. 33. Otvos, J. (1999) Clin. Cardiol. 22, II-21–II-27. 13. Bodzioch, M., Orso, E., Klucken, J., Langmann, T., Bottcher, A., Diederich, 34. Shepherd, J. (1994) Drugs 47, 1–10. W., Drobnik, W., Barlage, S., Buchler, C., Porsch-Ozcurumez, M., et al. (1999) 35. Jong, M. C., Hofker, M. H. & Havekes, L. M. (1999) Atheroscler. Thromb. Vasc. Nat. Genet. 22, 347–351. Biol. 19, 472–484. PHARMACOLOGY 14. Rust, S., Rosier, M., Funke, H., Real, J., Amoura, Z., Piette, J.-C., Deleuze, J.-F., 36. Staels, B., Dallongeville, J., Auwerx, J., Schoonjans, K., Leitersdorf, E. & Brewer, H. B., Duverger, N., Denefle, P., et al. (1999) Nat. Genet. 22, 352–355. Fruchart, J.-C. (1998) Circulation 98, 2088–2093. 15. Clee, S. M., Kastelein, J. J. P., Dam, M. V., Marcil, M., Roomp, K., Zwarts, 37. Leibowitz, M. D., Fievet, C., Hennuyer, N., Peinado-Onsurbe, J., Duez, H., K. Y., Collins, J. A., Roelants, R., Tamasawa, N., Stulc, T., et al. (2000) J. Clin. Berger, J., Cullinan, C. A., Sparrow, C. P., Baffie, J., Berger, G. D., et al. (2000) Invest. 106, 1263–1270. FEBS Lett. 473, 333–336. 16. Repa, J. J., Turley, S. D., Lobaccaro, J. M. A., Medina, J., Li, L., Lustig, K., 38. Peters, J. M., Lee, S. S. T., Li, W., Ward, J. M., Gavrilova, O., Everett, C., Shan, B., Heyman, R. A., Dietschy, J. M. & Mangelsdorf, D. J. (2000) Science Reitman, M. L., Hudson, L. D. & Gonzalez, F. J. (2000) Mol. Cell. Biol. 20, 289, 1524–1529. 5119–5128. 17. Brown, P. J., Smith-Oliver, T. A., Charifson, P. S., Tomkinson, N. C. O., Fivush, 39. Dietschy, J. M., Turley, S. D. & Spady, D. K. (1993) J. Lipid Res. 34, 1637–1659. A. M., Sternbach, D. D., Wade, L. E., Orband-Miller, L., Parks, D. J., 40. Suckling, K. E. & Jackson, B. (1993) Prog. Lipid Res. 32, 1–24. Blanchard, S. G., et al. (1997) Chem. Biol. 4, 909–918. 41. Rubins, H. B., Robins, S. J., Collins, D., Fye, C. L., Anderson, J. W., Elam, 18. Henke, B. R., Blanchard, S. G., Brackeen, M. F., Brown, K. K., Cobb, J. E., M. B., Faas, F. H., Linares, E., Schaefer, E. J., Schectman, G., et al. (1999) Collins, J. L., Harrington, W. W., Jr., Hashim, M. A., Hull-Ryde, E. A., Kaldor, N. Engl. J. Med. 341, 410–418. I., et al. (1998) J. Med. Chem. 41, 5020–5036. 42. Chinetti, G., Lestavel, S., Bocher, V., Remaley, A. T., Neve, B., Torra, I. P., 19. Hansen, B. C. (2000) in Obesity: Pathology and Therapy, eds. Lockwood, D. H. Teissier, E., Minnich, A., Jaye, M., Duverger, N., et al. (2001) Nat. Med. 7, & Heffner, T. G. (Springer-Verlag, Berlin), pp. 461–489. 53–58. 20. Otvos, J. D. (1997) in Handbook of Lipoprotein Testing, eds. Rifai, N., Warnick, G. & Dominiczak, M. H. (Am. Assoc. Clin. Chem. Press, Washington, DC), pp. 43. Chawla, A., Boisvert, W. A., Lee, C.-H., Laffitte, B. A., Barak, Y., Joseph, S. B., 497–508. Liao, D., Nagy, L., Edwards, P. A., Curtiss, L. K., et al. (2001) Mol. Cell. 7, 21. Berger, J., Leibowitz, M. D., Doebber, T. W., Elbrecht, A., Zhang, B., Zhou, 161–171. G., Biswas, C., Cullinan, C. A., Hayes, N. S., Li, Y., et al. (1999) J. Biol. Chem. 44. Staels, B., Ngoc, V.-D., Kosykh, V. A., Saladin, R., Fruchart, J.-C., Dallon- 274, 6718–6725. geville, J. & Auwerx, J. (1995) J. Clin. Invest. 95, 705–712. 22. Forman, B. M., Chen, J. & Evans, R. M. (1997) Proc. Natl. Acad. Sci. USA 94, 45. Hertz, R., Bishara-Shieban, J. & Bar-Tana, J. (1995) J. Biol. Chem. 270, 4312–4317. 13470–13475. 23. Venkateswaran, A., Laffitte, B. A., Joseph, S. B., Mak, P. A., Wilpitz, D. C., 46. Brown, P. J., Winegar, D. A., Plunket, K. D., Moore, L. B., Lewis, M. C., Edwards, P. A. & Tontonoz, P. (2000) Proc. Natl. Acad. Sci. USA 97, Wilson, J. G., Sundseth, S. S., Koble, C. S., Wu, Z., Chapman, J. M., et al. (1999) 12097–12102. (First Published October 17, 2000, 10.1073͞pnas.200367697) J. Med. Chem. 42, 3785–3788.

Oliver et al. PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 ͉ 5311 Downloaded by guest on September 28, 2021